F670 T103/203 transformer question

Question is about max headroom of this (sidechain amp input transformer). OK, it's 1+1:17+17, but sowter stated the max level at 50hz +16dBu or recalculated to volts it's a 13.8V p/p.
I really doubt that since ratio is 17 , this core (is that mumetal core or just can?) can deliver 234.6V p-p  . It must be that 16dBu is max at secondary. But that divided by 17 is 0.81V.... Looks too small . What you think about it?

heard rumors of some fairchilds coming with control input iron that was different from stock

like triad hs-14 or something weird.

the UTC is 1:9.7

the triad has a 1:14 or something weird

the pinout on the prints is indicative of the UTC transformer.

we could probably get Brian to wind a UTC version, since we have all the winding specs over in the Lab.

Thanks Chris. You didn't really answer my question but it's good to know about different versions of T103. But actually you answered because triad hs-14 (hs-3, hs-4) are low level transformers and they are definitely not capable to handle more than 10dbm. Since termination at the end of T103 is 100K I calculate 10dbm=90V p-p. Is that correct?
On the other side there is Sowter with +16dbu or 13.8V.... 13.8V at 100K or something else? That's the question now 

Except having dBm's and dBu's mixed up* (and p-p volts do not help), your first calculation is correct:
For 16dBu (4.9Vrms) at input, with 1:17 ratio, output level is 83Vrms =>234V p-p
Second one is not correct:
For 10dBu (2.49Vrms), if the transformer is 1:14 it's 35Vrms =>98Vp-p
A transformer can be optimised for such conditions, depends on the actual load. If the load is 100k nominal, power is actually only 69mW. I suspect the frequency response may not be very good though, unless the winding is highly sectioned, which is very costly. But the magnetic core is not a problem.

* dBm is one milliwatt power. If AND ONLY IF nominal impedance is 600 ohms, voltage is 0.775V. In RF, where nominal impedance is 50 ohms, voltage is 0.225V
Since the impedances are not specified and only voltage concerned, you should use dBu, i.e. dB related to a voltage reference of 0.775V

fortunately, we can compute the Exact gauss on the core from he winding info.
then figure in the pad...

so first, what frequency do you want to spec the xfmr at?

i will get the cross section

please wait  .....computing.....
here we go, this is how you get input level from winding data:

step a:

get the winding data from the fairchild symbol thread in the Lab-
winding data:

Core: UI31  15/32" stack

S1 red>F3 brn -500 turns #38- 4 Layers
S6 blk>F8 grn  - 4800 turns of #43- 21 Layers
S2>F4 - 500t #38 4L
S5>F7  - 4800t #43 - 21L


step b:

go to mag metals website and see if you can find specs for UI31-






click on Catalogs, not Products:



U will see a UI section, click on that,
choose metric, we will be doing physics with that number:




Yes Yes Yes!



really, thats all we need, we are donE with mag met,
but look, more interesting stuff if you click on the 31UI button,



depending on how you assemble the laminations, the magnetic properties will change slightly,

we call this stacking, it is in the B max equation, so lets take a look:

we need to tame this B max formula in regards to this xfmr:


Bmax = E * 100,000,000/(4.44 * k * cm^2 * f * N)

B max is flux at the top of the wave
E is AC Volts on the winding
100,000,000 is from maxwell'/faraday equations
k is stacking usually always 0.9 for laminated cores
cm^2 is the core cross section, (see that a smaller core generates more flux)
f is frequency
N is number of turns (see that as turns go up, flux comes down)

let us divide out some of the constants we already know for this Y-6431,

use k = 0.9 stacking, frequency = 20 hertz,

we need to knock down this silly 100,000,000 lines of flux biz:

100,000,000 / 4.44 constant = 22,522,522

22,522,522 / 0.9 stacking = 25,025,025

25,025,025 / 20 hertz = 1,251,251

so for any laminated transformer, if we want the B max at 20 hertz,

we can simply use this new formula  and punch in the cross section and turns, so

B max @ 20 hz = 1,251251/Cm^2 * N

since this 31UI core leg is 5/16", we would normally use 5/16"  *  5/16" to get the cross section

but we have a 15/32" stack, so to get the cross section,

15/32"  * 5/16"  =  0.14648 inches squared

convert to metric-

Answer: .14648 in² = 0.945 cm² ,  so we can dived that out:

1,251,251 / 0.945 = 1,324,075

and divide by turns and we are done:

1,324,075 / 960 = 1379

for the UTC Y-6431,      B max @ 20 hz = 1379 V ac or simply

B = 1379 V

lets approach it from another angle and see what numbers we get,

how much does the core weigh?
how much power will it take from a weight and material standpoint?

step a) gather data...
from mag met again...

so how do i get the core weight from that?

lets try the highlighted core material first, and see if we can get the weight...

15/32 core stack = how many inches?

oh, about 0.46875 i would hazzard to guess... 

we know all lams nowadays are always 0.014,

so just divide the stack height by lam thickness, we get

33.4 lams, so throw out the 0.4 for air space and other stuff, and we have:

33 lams

so now we have the answer in 212.3 lams per pound, so

33/212.3 = 0.1554 lbs

we might need that in grams, so we go to the online converter:

http://www.worldwidemetric.com/metcal.htm

easy as pie...



so

Core Weight: 70 grams or 0.07 kg

abbey road d enfer said:

Except having dBm's and dBu's mixed up* (and p-p volts do not help), your first calculation is correct:
For 16dBu (4.9Vrms) at input, with 1:17 ratio, output level is 83Vrms =>234V p-p
Second one is not correct:
For 10dBu (2.49Vrms), if the transformer is 1:14 it's 35Vrms =>98Vp-p
A transformer can be optimised for such conditions, depends on the actual load. If the load is 100k nominal, power is actually only 69mW. I suspect the frequency response may not be very good though, unless the winding is highly sectioned, which is very costly. But the magnetic core is not a problem.

* dBm is one milliwatt power. If AND ONLY IF nominal impedance is 600 ohms, voltage is 0.775V. In RF, where nominal impedance is 50 ohms, voltage is 0.225V
Since the impedances are not specified and only voltage concerned, you should use dBu, i.e. dB related to a voltage reference of 0.775V

Click to expand...

It's possible that I mixed up something but Triad stated 10dBm not me. So I just tried to calculate voltage saturating point based on that. I presume that they used 600ohm as reference because of audio transformer , so in that case voltage portion of power should be  2.49V at 10dBm  . At this point same result is using dBu. Sounds reasonable, 98V p-p but why then they call it "low level transformer"???
On the other side 243V p-p with sowter 8345 sounds a bit weird to me. Unfortunately they didn't stated 16dbu at input , just "max level at 50hz" . Max level is at sec. ? am I right?

Chris you are legend  . What to say? But it takes time to follow you ... I have a time but still can't figure exact number... 

i am having a rough time also. so i am explaining it to myself online  :
let us see if there is any light at the end of this tunnel
so far we have

1) B = 1379 V
2) core weight = 70 g or 0.07 kg

now we have one last method
we did the formula for B max
we did the weight method,
now the volume method,

then we will convert all 3 to db and see what they meeeeeean...

Area Product

area product is the product of the window area in the middle of our UI core
and the cross sectional area

for some reason, the xfmr engineers like it, (that means it saves time) 

Ap = Wa (window area)  * Ac (cross section)

both Wa and Ac are in centimeters squared, so the Ap units will be  <cm.^4>

so we have all the dims already for this lam
that means we have enuff data to use the formula

but let us do the Area Product in inches first, just in case we need it later,

0.14648 inches squared  (cross section from above posts)

and we also need the window area in inches,




0.9375 * 0.3125 = 0.2929 inches squared

so our Ap in inches is

0.2929 in^2 * 0.14648 in^2 =

0.0429 in^4

now metric:

0.9375 * 0.3125 = 0.2929 inches squared =  1.89 cm^2

we have our cross section already from above:    0.945030 cm^2

so just multiply the window area by that and we have our Area Product for this transformer and this lam:

0.945030 cm^2 * 1.89 cm^2 = 1.786 cm^4

so

Ap = 1.786 cm^4

and

Ap =0.0429 in^4

now there is a Volume to Area Product relationship in transformers,
it goes something like this:

V = Kv * Ap^0.75

where Kv is a constant related to core configuration,

for laminations, Kv = 19.7,

so V = 19.7 * 1.786^3/4

V = 19.7 * root 4(1.786^3)

V = 19.7 * root 4(5.697)  now just take sq root twice of 5.697

V = 19.7 * 1.545 =30.4 cm^3


so we have three four different programs to work:

1) B = 1379 V
2) core weight = 70 g or 0.07 kg
3) volume = 30.4 cm^3
4) Area Product = Ap = 1.786 cm^4

and, those lams are actually the same as the UTC A-10, not the neve, Doh! 

and the stack is about the same, so we can also use the A-10,

just read the db reading on the can, although rumor has it that it will take a bit more?

tomorrow we will conclude all this mess and come up with some solid answers, later.

tomorrow we will conclude all this mess and come up with some solid answers, later.

Click to expand...

Thanks, that will be great  ;D

Moby said:

It's possible that I mixed up something but Triad stated 10dBm not me.

Click to expand...

Ok, so Triad used dBm's because at the time there were no dBu's or dBv's. But today, as soon as we're talking about voltage, we should correct to dBu's.

So I just tried to calculate voltage saturating point based on that. I presume that they used 600ohm as reference because of audio transformer , so in that case voltage portion of power should be  2.49V at 10dBm  . At this point same result is using dBu. Sounds reasonable, 98V p-p but why then they call it "low level transformer"???

Click to expand...

Because technically, for a transformer manufacturer, it's a low POWER transformer.

On the other side 243V p-p with sowter 8345 sounds a bit weird to me.

Click to expand...

What do you think is weird? A standard output xfmr can withstand voltages of more than 100Vrms at 1kHz. The main limitation in level-handling is saturation of the core, which increases as an inverse function of square of frequency. A transformer that accepts 1V @20Hz will accept 100V @200Hz

Unfortunately they didn't stated 16dbu at input , just "max level at 50hz" .

Click to expand...

That's a good indication of the typical operating level. For applications that don't require a particularly low LF extension, I would think that's the recommended max level, with a nominal operating level of about 10-12dB below; interestingly, that works out as +4dBu. For high-end applications, I would decrease these values by about 15dB to make the unit comfortable at 20Hz.

What do you think is weird? A standard output xfmr can withstand voltages of more than 100Vrms at 1kHz. The main limitation in level-handling is saturation of the core, which increases as an inverse function of square of frequency. A transformer that accepts 1V @20Hz will accept 100V @200Hz

Click to expand...

Maybe not weird for steel lams but for mumetal sounded a bit weird. I used to see mic input mumetal lams transformers and they can't stand high levels. OK, maybe this one can 

they used Supermalloy  in there, i believe,

have to calculate 960 turns = 1 henry on a 31UI...


and the input power is always measured on the 600 ohm side of the transformer.

otherwise you have to say "20 db @ 39 k ohms" or something weird like that.

remember that the db will hopefully be the same on both sides of the xfmr!

Thanks Chris, I appreciate your efforts 

dug up this olf file on an A-10 dissection:




so the Y-6431 probably has a few more lams, as the A-10 has a smaller lam stack
we were close, look at the weight  we calculated for the 6431:

33/212.3 = 0.1554 lbs

almost matches the actual 0.15 lbs weight i measured on a real good scale. 

kevin over at k and k audio told me i messed up the ha100-x specs,
says iy has a few more lams than the A-10.

this would put the Y-6431 right there with the HA-100X.

the Y-6431 has 4 cans, same with the HA-100X

both have about 33 lams, and i found some old inductance measurements for the HA-100X

they match the Y-6431 very closely, the same machine was used to measure both,
so since the 100x and the 6431 both have about the  same turns and the henries are the same,
this means that the core material is the same, so we have just nailed down the core,

this means that the HA-100X will have about the same input level as the Y-6431,

the A-10 comes in at 15 db
the HA-100X + 16 db

so i am guessing the Y-6431 will handle 16 db also.


here is a chart...



16dbm (excuse me for using db like an ingnoramous) = 40 milliwatts

so where are we?

1) B = 1379 V
2) core weight = 70 g or 0.07 kg
3) volume = 30.4 cm^3

well, now that we know the core material is Supermalloy, we can solve 1)

Supermalloy laminations will only tolerate about a maximum flux of  5 kilo gauss,

since from our B max formula we know that if we hold frequency, turns, and core constant, the flux will then be directly proportional to the last unknown variable, (big shots call this a differential) ,which is  AC volts rms, V, or E in some books.

so if B = 1379 V, and B = 5 kilo gauss max for supermalloy, we can calculate the max voltage:

5,000 gauss / 1379 = 3.6258

how many db is that?


go here to convert volts rms to dbm at any impedance..

http://www.raltron.com/cust/tools/voltage_to_dbm.asp



so 13 1/2 db, if you figure UTC cheated and rated it at 30 hz, you could get 16 dbm

so our homebrew calculations look ok for this transformer so far...  ;D

16dbm (excuse me for using db like an ingnoramous) = 40 milliwatts

Click to expand...

CJ, the use of dBm in POWER calculations is perfectly justified, and you know it

Thanks Chris I enjoyed reading and learning. Also thanks for "young man" , is sounds so great when you are 42 
Anyway, in the meantime, I decided to contact Brian too, so he confirmed spec for 8345 (T103 replica).

8345 will take +16 dBu at the input (both primary windings in series) at 50
Hz and above.  The ratio is 1+1 : 17+17

All these transformers were made as a result of measuring the originals and
modifications made after testing several complete units so we are very
confident about their performance in the original design.

Click to expand...

Cool!

ask him which xfmr he modeled after, or is it on the web site?  :

i re did the above posts, cleaned them up and deleted one, so it reads a lot easier now, threw out the stupid spread sheet approach and simplified the b max,

I have both an HA100X and a Y-6431 already here, so I can sweep the hell out of them all day long,

and, I had the B max formula wrong, it's 100,000,000, not 10,000,000,

damn maxwell!

or faraday!

i wanna kill em both for using such large numbers!
:-* :'( :-\ :-X 8) ??? ??? ??? ??? ???

so the dbm looks like 13.4, not 31!

so the motto is: don't believe everything you read on the net, at least until cj is done editing...




off hand, i would say that the HA100-X would be the perfect drop in replacement for the Y-6431.

same lams, same number of cans, similar shape, same turns, almost the same dcr,

only difference is the HA100-X has all those taps brought ought.

the Y-6431 was probably a special order HA100-X, only with no taps.

ok, this is really cool, i just read, that:

in order to double VA for a laminated transformer, you only need 68% percent more lams, not 100%.
so just figured out that is why B max is specified for a Square Stack ONLY!

since the current density J is fixed for a given window, the core grows at a different rate than the copper as the stack height is increased.

this means that the steel/copper ratio is a factor in determining VAR and all that power stuff, i am trying to wrap my head around this as we speak,

if you figure this fact into our 13.4 dbm calculation, that 15/32 core will really take more than we thought.

oh, cool, maxwell and faraday are already dead, checked wikpedia, so i feel better noew,
i don't have to hire keef richards to take them out.  :'(



ok, i am doing my part, who is going to handle the 670 pad calculation?

because when it comes to pads, i don't know rat shit from rice krispies, 

where is NY Dave when you need him>
he is the Pad King!

probably drinkin a cold one at aes, lucky bastard!  ;D